Currently, in multiple-input multiple-output (MIMO) communication, a closed-loop precoding transmission solution that is used in a low-speed scenario and based on channel estimation may provide a quite high array gain. However, when a moving speed of a user increases or when there is a hardware error, a direction of a beam transmitted by a base station is prone to not matching the user, causing a performance loss. Especially when there are a quite large quantity of antennas, a performance loss is more severe. Generally, a base station performs data transmission according to a beam-unrelated parameter of user equipment (UE), which cannot change a status or quality of a link. As a result, an application of a large-scale antenna system in a frequency division duplex (FDD) scenario is greatly restricted.
Based on the foregoing description, in the prior art, a direction of a beam is prone to not matching a user, which causes a relatively large performance loss; especially when there are a quite large quantity of antennas, there is no appropriate data transmission solution in an FDD scenario.